Background and aim
Metastasis-associated lung adenocarcinoma transcript 1 (MALAT1) is implicated in cancer-related gene regulation and is a target for transcriptional modulation in lung adenocarcinoma. CRISPR interference (CRISPRi) guide selection must balance predicted efficacy, local genomic context, sequence feasibility and genome-wide specificity. We developed an uncertainty-aware framework to prioritise MALAT1 CRISPRi guides in A549 cells and inform reproducible cell-physiology experiments.
Methods
Eighty-six guides within ±500 bp of the MALAT1 transcription-start anchor were assessed using six criteria spanning four domains: target position and chromatin accessibility; local transcript and promoter overlap; sequence feasibility; and predicted CRISPRi specificity. A customised stochastic multicriteria acceptability analysis type-2 framework sampled hierarchical uncertainty in criterion weights over 30,000 Monte Carlo iterations for each of three preference profiles. A profile-free global-weight stress test used 50,000 iterations. Position-model, sequence-quality, off-target-model, leave-one-domain-out, convergence and random-seed analyses assessed robustness.
Results
Candidates 0046, 0047 and 0048 formed a robust core. Their minimum probabilities of ranking in the top five across the three profiles were 0.980, 0.919 and 0.601, respectively; corresponding means were 0.993, 0.969 and 0.807. Under profile-free globally sampled weights, their top-five probabilities were 0.817, 0.741 and 0.519. Candidate 0045 remained competitive but shifted from rank 3 to 29 across off-target models. In one-at-a-time analysis, weight uncertainty produced the largest single mean absolute rank shift (1.49 ranks), compared with off-target-specificity modelling (0.90) and local-context uncertainty (0.64).
Conclusion
Integrating predicted efficacy context and specificity with explicit uncertainty identified candidates that remained competitive under alternative decision assumptions. By distinguishing consistently high-ranked guides from context-sensitive alternatives, the framework can support guide panels for cell-physiology experiments testing how MALAT1 repression affects transcriptional state, proliferation, migration and stress responses. It can also separate guide-selection uncertainty from downstream biological variability, strengthening interpretation and reproducibility.
Ethics
This study used no human participants, identifiable human data, human tissue or animals; research ethics approval was not required.
Body word count: 300 • Candidate set size: 86 • No biological sample size or inferential p-value applies to this in silico study.